List your company

Mathematical Reasoning on GSM8K (Accuracy, Token count)

96.6AccuracyVanilla
92.4493.5294.695.68Apr 8, 2026

Evaluation results

MethodLinks
VanillaBackbone=Qwen3-32B2026.0496.61,707
VanillaBackbone=Qwen3-14B2026.0496.21,672
DTSRBackbone=Qwen3-14B2026.0496.2849
VanillaBackbone=Qwen3-8B2026.04962,208
DTSRBackbone=Qwen3-32B2026.0496849
NoWAITBackbone=Qwen3-14B2026.0495.81,133
NoWAITBackbone=Qwen3-32B2026.0495.81,212
DEERBackbone=Qwen3-32B2026.0495.5979
DEERBackbone=Qwen3-14B2026.0495.3896
DEERBackbone=Qwen3-8B2026.0495.1963
DTSRBackbone=Qwen3-8B2026.0495.11,152
NoThinkingBackbone=Qwen3-14B2026.0495.1278
NoWAITBackbone=Qwen3-8B2026.0494.91,307
NoThinkingBackbone=Qwen3-32B2026.0494.8283
NoThinkingBackbone=Qwen3-8B2026.0492.6303